Colloquium: Urh Krzic (C&O)

25 October 2019 14:30 - Location: Lecture Room K, Faculty of Aerospace Engineering, Kluyverweg 1, Delft.

Aircraft Maintenance Analytics: A data-driven approach with real-time implementation.

This research develops prognostic health monitoring for an aircraft system and integrates the analytical information into aircraft maintenance organization.
Several suitable systems are identified based on analyzing aircraft delays, data availability and failure mechanisms. The aircraft bleed air distribution system was chosen as the ideal candidate for health monitoring. Relevant aircraft and organizational data was extracted for this system and normal and failure behavior identified by exploring and processing the data. Real-time fault alerts were developed, based on the regular extraction, parsing and processing of aircraft data. A user interface was created to host these alerts and data.
This interface was also enhanced with additional to help with fault diagnosis and troubleshooting. Prognostics were developed to improve decision-making ability. The predictive information generated consists of a trend visualization and basic forecasting module, and a Random Forest Remaining-Useful-Life estimation module.